Medical field sees big potential for AI

S1: It's time for Midday Edition on Kpbs. Today , we are talking about the latest uses of fast advancing AI technologies. I'm Jade Hindman. Here's to conversations that keep you informed , inspired and make you think. We'll explore how local hospitals are using artificial intelligence.

S2: It's being used and a lot of these pilots , where it's used in small groups of physicians at each of the health systems here in San Diego , but it isn't being widely embraced.

S1: Plus , the latest discoveries in how the technology could help patients. And one local journalists let artificial intelligence run her life for a week. She'll join the conversation to tell us how that turned out. That's ahead on Midday Edition. Artificial intelligence is finding its way into just about every field imaginable , but especially the medical field where real world benefits to physicians and patients alike are being explored. And a recent grant awarded to the Scripps Research Translational Institute by the National Institute of Health will help further the fast developing relationship between AI and medicine. Dr. Eric Topol is founder of the Scripps Research Translational Institute and joins us now. Dr. Topol , thanks for being here.

S2: Thanks , J. Always good to be with you.

S1: So first , congratulations on receiving a very large NIH grant , nearly $50 million.

S2: We use digital sensors and genomics and AI to understand the uniqueness of each person and promote their health. And also , of course , training the future physicians and scientists that will help do the same. That is a whole idea of precision medicine and individualized attention. So things like how we screen for cancer and how we prevent diseases that otherwise people are high risk to manifest.

S1:

S2: We are a life science capital of the country in the world here. So many great things are happening at the different institutions on the Mesa , no less , of course. A lot of the life science industry in the region. So we're really privileged that we here in San Diego have some of the most extraordinary science and support by the NIH.

S1: You know , a big area that the institute has focused on in recent years , as you mentioned , is AI.

S2: That's the big project that goes for the next seven years. We call it multimodal AI because what it means is bringing all the layers of data about any given person. So , for example , you know , we envision that hospitals will largely be different in the years ahead where only the sickest people in the intensive care unit or needing an operation or emergency department visit , but otherwise all the other patients will be at home and they'll have sensors and they'll have all their data , the layers of data , like their electronic health records , their environment and their sensors , all with the medical knowledge , everything that's ever been published relevant to that person , continuously integrated in process and providing surveillance so that the person doesn't need to go to the hospital , doesn't have any compromise. By the same token , the virtual health assistant where you would be coached if you're interested , to not have a condition that you're at risk , or if you have a condition to try to manage it far better than we do today. So these are some of the projects that we have as part of the Big Grant. You know.

S1: When we last spoke , you were at a at a breaking point with AI and medicine. Since then , experts in the field have been discussing the possibilities of ChatGPT for and its potential applications.

S2: This GPT four , which was a jump on its predecessor , and then many more of these large language models , also known as generative AI to come the capabilities are just extraordinary in medicine and health care. So it's not just the fact that we are already seeing that you don't have to sit at a keyboard as a doctor or a nurse , but you can just have the conversation made into not just great notes , but all the other things that you'd want to do in terms of ordering tests or prescriptions or follow up appointments and nudging patients about the things that that they need to attend to. But it's also about being the front door for patients to be able to ask questions , concerns about their health , how to how to preempt the need for physicians as a front door where it's hard to access them. We're already seeing how many examples that the questions can be answered as well , or even better in many cases with these large language models. So the future here , no less the things. I talked about , like the gutting of hospitals and virtual health coaches and so many other things will eventually take hold. And that's why it's so exciting. We've never had these capabilities before.

S1: I'm curious , with the advancement of AI.

S2: And so the average time with a patient is less than seven minutes in a clinic visit. That's not adequate at all. So the whole idea is to get the gift of time. And as you've alluded to , the idea is that when you see a patient , all that information is is basically distilled and crystallized. So you don't have to go searching through pages and pages of the person's electronic chart and up to date with respect to all the medical knowledge pertaining to the patient. So you don't have to be necessarily reading all the literature or looking it up after you see the patient. And so the help in making more accuracy in diagnosis and also reducing the need for visits because some of that can be accomplished through these large language models. So that gets us to patients having more autonomy , doctors and nurses being less compressed for less squeezed for time and giving back this gift of time to patients so that the relationship , the patient doctor relationship can be restored to the way it used to be many decades ago and potentially even supersede that , which would be extraordinary.

S1: Also with this , the large language models would this type of AI be able to sort of compare notes from the specialists that one is seeing , put it all together and give a diagnosis much faster ? Yeah.

S2: You're touching on something that's really pretty remarkable when there you not only can get a second opinion or a first opinion , but you can then audit that opinion by just having a new conversation. And so , yes , that we have a serious problem with diagnostic accuracy in this country with more than 12 million serious diagnostic errors per year. And so the whole idea is that now we have a kind of auxiliary pathway to get information to help promote diagnostic accuracy. And the other thing to note about it , it's not just about text and speech. It's also about images. So you'll be able to take your x ray or your past slide and put it into one of these models to get a second opinion. So the ability to get of course , so much of this has to be validated and there's everyone knows about this hallucination issue where the chat bot can make stuff up , which obviously we can't tolerate that to any degree. But this is some opportunities that we've never seen before , and the high on the list is getting far more accurate diagnoses.

S1: Dr. Topol I want to bring up this latest article from the New York Times. The headline reads , I may Someday work medical Miracles. For now , it helps do paper work.

S2: It's making these great notes and taking care of all the other mechanical things , like I mentioned , making follow appointments , doing pre authorization to an insurance company so the doctors don't have to deal with that burden and so many other things to get towards keyboard liberation. And one of the doctors said it takes me out 20 minutes to what it did take two hours. But the other thing that was really striking about this , Gene , was that the patients really loved it because it gave them information that they'd never had before. It gave them the full note , the full conversation , and pointed out that 80% of what's discussed at a visit is forgotten by the patient. So to have a record of everything that was discussed and to be able to do this trust and verify , whereby when the note is made , if there's any questions. It then will link right to the conversation where that point was being made. So unlike the use of these GPT large language models for hallucination , the good part here is everything can be fully indexed to what was said. There's complete transparency , if you will , and trust. So this is a big movement going on right now and throughout the country will be seeing the use of notes based on the conversation rather than clicking away. And it brings back face to face contact listening to patients and that humanity that's been largely missing the the human human bond that is essential for the practice of medicine.

S1:

S2: And remember , you know , 50% of doctors are below average , right ? Unfortunately , that's true. And that means part of their diagnosis or their communication or whatever that you're you're looking at in terms of how doctors function , there's a lot who are below average. And so while large language models will never be perfect , they they are showing us already here in 2023 some capabilities that are pretty extraordinary. And so the point here is that you have humans in the loop. You still have doctors providing oversight , but you have in many ways another checkpoint , another pathway to get information , to help make diagnoses , to make lives easier for patients and physicians giving more autonomy to patients. So , for example , today , people can do a Google search and that information they get about their health is oftentimes very misleading. And now they can , you know , search ChatGPT or GPT four. And there will be other ways soon that are even more powerful than what we have today. So it's never going to be perfect. But look , look at the standards we have today with respect to how we do searches and our interactions with doctors that are not always getting the right information. So I think it's it's going to be another important source. And most people in the years ahead will be using AI as part of their health care.

S1: You know , there's there's often a stark divide in the kinds of medical outcomes for people of color compared to white patients.

S2: So being on the lookout for what are the populations where the inputs are available are critical. So interrogating these models for the the inputs is essential. And then all the other aspects of being able to assure fairness , lack of bias , it's typically not in the algorithm itself , but rather what's put in. It's embedded in our culture. But in medicine , this is really important. We've seen flagrant examples that if we don't have broad , diverse population input across all ancestries , all ethnicities , that we can have flawed outputs and interpretation. So this requires a lot more attention. But one thing we can say is that this is basically software. It's inexpensive and it can be applicable throughout the world. And so we've already seen the examples of , let's say , smartphones that have an ultrasound probe attachment being used in the hinterlands of Africa and India and being able to do imaging across any part of the body without even people trained to do ultrasound imaging , no less AI interpretation. So there's a lot of opportunities here where I could bridge the inequalities and we shouldn't lose sight of that. There's there's many examples and already around the world there is a lot of software to diagnose TB accurately in places of low and middle income countries that didn't have resources. And we'll be seeing much more of that in the years ahead.

S1:

S2: So we've talked a little bit about the AI supporting humans , but we haven't yet gotten to just now AI's interrogating one another or exchanging information. So there's this fear , of course , that AIS are going to somehow take over. I think that's greatly exaggerated , but we are going to see , like , for example , in the health care world we're already seeing. So instead of doctors having to fill out all these pre authorization forms for a patient to get a particular medication or a test , it's done by AI. And then the insurance company has an AI that goes back and forth with it. There's no humans in the loop there. And until if indeed it becomes necessary , we'll be seeing that sort of thing across the board.

S1: Do you know if there's any health care systems here in San Diego using AI ? It seems like it takes forever to get an appointment , and I would imagine the use of AI might expedite these things.

S2: But it isn't being widely embraced. And I think , you know , that's going to happen more over the next year or two , where you see the example you've given where maybe you don't even need a physical appointment , maybe you just need front door help from one of these generative AIS , but also triage so that you can get an appointment much easier. The average person in the country take three weeks to get an appointment with their primary care doctor. That's not adequate and forces many people to go wind up going into an emergency room , which is not a good choice in many , not just for cost , but also for wait times and what can happen in emergency rooms if you're not acutely ill. So we've got to do better. And you're you're bringing up one of the areas where this wait to get to a doctor. It's unacceptable. And I'm sure particularly here in San Diego , we can do much better.

S1: You spoke with former Vice President Al Gore about the intersection of AI and climate change.

S2: It was really exhilarating discussion. We had a lot of people identify Al Gore as one of our world leaders in climate change , but he's also quite an authority on AI , And he points out that we're heading to the point where we're not going to be the apex lifeform in terms of intelligence , because the intelligence that we've already seen exhibited by these large language models in many situations exceeds that of human intelligence. It's not just that these models are some kind of stochastic parrot , but they actually are coming up with cause and effect and doing things that we know represents more intelligence than we ever seen by machines. But the other thing is that because the amount of graphic processing units , these computing units that are required to train these models is enormous , you know , to tens of thousands of GPUs to train GPT four and others. The amount of electricity that's consumed and the amount of water that's needed to keep these computing facilities cooled is enormous. So it could make climate change worse. And one of the things he pointed out also is he thinks he's sure that I will help us come up with solutions for the climate crisis. We haven't seen so many of those yet from AI , but as he asserts , it's getting more intelligent all the time.

S1: I've been speaking with Dr. Eric Topol , director of the Scripps Research Translational Institute in La Jolla. And Dr. Topol , as always , thanks for talking with us today.

S2: Oh , thanks so much.

S1: How do you feel about doctors using artificial intelligence ? Are the possibilities exciting or does it all make you a bit nervous ? Give us a call at (619) 452-0228. Leave a message or you can email us at midday at Kpbs. Coming up , the conversation continues as we look ahead to how I could soon be used.

S3: You can improve diagnostic effectiveness with the help of a superintelligent digital sidekick. And that's been the vision for artificial intelligence in every field for a long time.

S1: You're listening to Kpbs Midday Edition. Welcome back. You're listening to Kpbs Midday Edition. I'm Jade Hindman. Earlier in the show , we heard from a renowned medical expert on the potential impacts I could have in the field of medicine. The future of AI in health care may look bright , but our next guest argues for now , the reality of things is a little more grounded. Steve Lohr is a reporter for The New York Times who covers technology and its impact on health care. He joins us now. Steve , thanks for being here.

S3: Thanks for inviting me.

S1: So I in health care , it's it's not just a pipe dream. It's currently being used. Now. Can you tell us a bit about how it's being used to aid physicians in the field ? Yeah.

S3: One of the is one of the things it's being used to is to to lessen this horrible documentation burden that doctors and nurses have of typing up their reports for billing and for patient records. And that can take hours in a day and sometimes often lasts into the evening , particularly for primary care physicians. And they talk about that as pajama time , you know , keeping up with the digital paperwork. And so what what the latest degenerative AI developments have done have made it possible to just have your smartphone turn on an app. It listens to the conversation between the doctor and the patient. It makes a summary both for the patient at a kind of simplified level , so called fourth grade reading level. And then for the doctor , it it makes a patient report that complies with both treatment and billing requirements. And it seems to be pretty effective. It's it's improved a lot in the last year or so.

S1:

S3: I mean , there's other things that artificial intelligence has been used for for a while. I mean , what what we're talking about here with this generating summaries and the like and the sort of chat , GPT style is what's called generative. There's also predictive AI , which has been around for for a while and been developing for a while. And there are the federal Food and Drug Administration has approved a number of AI algorithms for things like reviewing and spotting cell clusters in in X-rays and radiology. That's less the sort of that's an assist rather than this kind of clever assistant that's being used. And and there are other uses. I mean , there's one of the things people are using it for is , is to communicate with patients to help them write sympathetic letter notes and messages to patients and communicate with them. And they've used ChatGPT as a way to kind of get that started. It's also being used for other kinds of messaging as well. And the people the broad view is that it will again transform things that every doctor will have this super intelligent guardian angel looking over their shoulder with giving them recommendations and and suggestions that and that guardian angel will have digested all of medical literature and will be up to date. And , you know , that's the sort of this intelligent assistant that will help them make diagnostic decisions in the future. It's , you know , it's not here now. It's coming. There's a debate about how quickly it will come. And and you know , what what hurdles it will face along the way in terms of regulation , legal requirements and the like. Medicine is not a free range for experimentation and innovation. You know , you don't do beta tests or you shouldn't in medicine.

S1: Even though they call it medical practice. But yes. Exactly.

S4: Exactly.

S1: I mean , it sounds like AI is really solving a lot of problems for people in the medical field. One of those things I would imagine is burnout.

S3: And if you can take 2 or 3 hours in a day out of of this kind of clerical work that doctors have to do , that reduces the stress considerably. And when surveys are done of doctors and nurses , one of their prime complaints is all the all the work on the electronic health record that they have to do in all the typing. One of the people I talked to for this story developed some of this. His father was a physician. He said his dad got out of the field. He just couldn't type fast enough anymore.

S1:

S3: It's still early for this kind of thing. And doctors , their technology has been used before. The automated speech recognition for transcription. There's been things like that that doctors have tried along the way. But if you go to a , you know , go to most doctors these days , you're still , you know , they're typing , right ? They're often at a computer. That isn't going to change overnight , but it is beginning to , you know , to move into physicians offices. But it's fairly early days for the. Kind of kind of fully automated summary of a conversation and and and , you know , reports the kind of , you know , digital scribe that I've described.

S1: Is there any hesitance to adopt this kind of technology within the field ? Yeah.

S3: I mean , you know , people have to check it first to even though it writes the note , you have to look over it. And there's this technology is kind of now famous for sort of making things up. So called hallucinations and hallucinations don't work in , you know , in medicine or shouldn't. So there is a lot of that kind of concern at the outset. And , you know , before people have tried it and many people have been through previous iterations of great promises about how technology is going to transform health care. Right. You know , it's like one of life's first principles. Everything takes longer and costs more than originally advertised. But there is this sense with current state of play of this technology that that it is sort of a watershed and this technology could really make a dent in that that digital paperwork burden that doctors and nurses face.

S1: So , Steve , we often hear about a lack of human connection in medicine as a result of these increasing patient loads and so on.

S3: And some of the people that have used this technology have been real enthusiasts. I mean , they say , you know , I'm finally 100% present for my patients. And that's you know , it's funny. The technology might be the tool to make the sort of the health care process more human. That is certainly I mean , one of the areas for real optimism.

S1:

S3: So that's been and there are studies that show patients forget , you know , 80% of what's said in the doctor's office , right. Over the course of a of a few days. And so you had this you had this record that is fairly easily searchable. And they're called medical moments for covers , medications , wellness suggestions and the like. That the person can is a little icon. They can tap on it and they can listen to that portion of the conversation again to remind them so they like , you know , doctors like that. In terms of of patient care.

S1: We've been zeroing in on doctors. But let's talk a bit more about patients. How might patients benefit from AI in health care , better.

S3: Communication , constant kind of updates and tailored care potentially. I mean , you know , so much of health care spending and people suffering is based on kind of chronic illnesses with heart disease , cancer , that sort of thing. And these are , you know , the diabetes , asthma , these are conditions you manage over time. So the traditional way most people deal with health care is you go to the doctor when something's wrong , as opposed to a more continuous communication where , you know , you can be suggested to make sure you take your medicine and follow up by various things. You can it can be potentially a much more personalized treatment for for patients. So that is opposed to going to the doctor a couple of times a year. And you know here's this and then you leave and come back only when there's a real problem.

S1: We touched on this earlier , but what kind of advances do they hope I will be able to make in medicine in the future ? In the.

S3: Future , I mean , the big target is diagnosis That's not there yet. But again , this you know , you can improve diagnostic effectiveness with the help of a digital super intelligent digital sidekick. And that's been the vision for artificial intelligence in every field for a long time. But it starts to feel particularly this , you know , chat GPT style , generative artificial intelligence that that actually this may really open the door to to to that world.

S1:

S3: So so there's that. And then it's a highly regulated field. It's , you know , it's strained already. And introducing anything new is many people will feel a doctors and nurses , oh , this is another new thing. So it has to prove itself that it really does save time and effectiveness. And then there's all the regulatory and legal concerns that that are legitimately there. I mean , nobody these days is taking , you know , a person your health record with all your background history and stuffing it into ChatGPT and asking it for a diagnosis. I mean , that's probably a violation of law.

S5: Of Hipa to. Do.

S4: Do. To do that ? Yes. Yeah.

S3: So it is a huge high stakes field that's under strain already. So technology is great if it can prove that it works and really delivers the benefit. But you know , you kind of win those battles one by one. And doctors , you know , nobody is that I was aware of is forcing doctors to do this. This is something that's made available to them through they're part of a medical center system or something. And so they're trying it out. And , you know , you have to kind of win hearts and minds.

S1: There's also a lot of fear that I will be a job killer in certain job sectors.

S3: And this is not going to this is not going to knock people. And then in any near term , you know , within five years , it's not going to eliminate jobs except in medical transcription , which has been a declining field for a while already because voice recognition has eaten into that. But otherwise , you know , if you can help doctors and nurses do less tedious tasks that to free up some more time in their day that's you know that's a big win that that doesn't it doesn't look like it's going to be a job killer in that field in the way that in some fields it may well be.

S1: I've been speaking with New York Times tech reporter Steve Lohr. Steve , thanks so much for joining us.

S3: Thanks for having me , JD Appreciate it.

S1: Coming up , one local journalist allowed AI to run her life for a week and she lives to tell us about it.

S6: Even though ChatGPT is generating a lot of buzz right now for feeling more on the intelligence side than the artificial side of artificial intelligence , there is still this sense that it doesn't totally know what it's talking about.

S1: Kpbs Midday Edition is back after the break. Welcome back. You're listening to Kpbs Midday Edition. I'm Jade Heinemann. For our next segment , we're going to go in a slightly different direction. We've been talking about the use of AI , particularly in the field of medicine here on Midday Edition today. But as our next guest can attest , artificial intelligence can have plenty of everyday applications. Just ask San Diego magazine associate editor Amelia Rodriguez. She let AI run her life for an entire week , and she's here now to give us her takeaways. Amelia , welcome. Hello. I have so many questions here. You let AI run your life for a week. I mean , that seems like a pretty big leap of faith.

S6: I think anything that runs the risk of disrupting your entire industry , it's something that you pay attention to. So a lot of my colleagues were kind of playing around with AI for a while. You know , we had our digital editor , Nicole Monaco was meal planning with ChatGPT , our CEO , Claire Johnson. Um , let AI generate headshots for her , um , that she didn't use anywhere. She was just curious about how well the , the technology could recreate her face. And so I became really curious about what else I can do. Originally it started as sort of a story that was about meal planning with ChatGPT , but it really took on a life of its own. I kept finding different ways to sort of hand over the reins to artificial intelligence , and it grew into what you see on our site now. Wow.

S1: Wow. I mean , and of course , you know , I would imagine some people might be a little uncomfortable with this idea. Did you have any reservations ? Definitely.

S6: I think that , for one , the aspect that was most uncomfortable for me was definitely I created a bumble BFF account. My photos were AI generated , my bio was generated , and all of the conversations that I had with folks , all of my responses were generated. And that's a relatively low stakes use of AI for human interaction. But , you know , as Rick Klein was just writing about the film , her for The New York Times and kind of talking about what might happen when , if and when I becomes our companions and we are like in social relationships with artificial intelligence. And I think that me having these conversations with people using ChatGPT brings up some of those fears for me and probably for other people. You know , the the immediate question is , are we going to be able to tell when we're talking to someone who who isn't real ? And I think that was my fear. I didn't want to to sort of trigger that fear for people. And that was my my biggest reservation going in.

S1: It sounds so fascinating yet so creepy at the same time. I mean , are.

S6: People so yeah.

S1:

S6: I think that people are becoming aware of ChatGPT , especially its potential as a useful tool. Certainly prompts can be refined. I talk at the end of the article about some of the ways that people can use ChatGPT to streamline their lives. You know , meal planning , exercise , planning and it can certainly make things easier for people who are first sort of stepping into those realms. But I think that even though ChatGPT is generating a lot of buzz right now for feeling more on the intelligence side than the artificial side of artificial intelligence , there is still this sense that at least when I use it , it doesn't totally know what it's talking about. I didn't feel convinced by anything that GPT was telling me to do , and I didn't see like a huge improvement in my life from letting it run it for a week. And so I think that we're not at a point where people are really relying on artificial intelligence to shape their decisions.

S1:

S6: I use ChatGPT a lot. It was planning my meals , it was planning my workouts and it was writing those responses to people on Bumble BFF. It was also it wrote a text to my girlfriend. So I was using it to interact with folks. And then I also allowed artificial intelligence to plan my outfits. So there's an app called Style DNA where you can upload all of your clothing. I spent hours photographing everything that I owned , and then it generated outfits based on that.

S1:

S6: I think if I wasn't doing an experiment of this nature , that's not something that I would do all the time. But I do think that if the app got smarter and was able to input more of my style preferences , if I was able to tell the AI what the weather was today , um , I would find it more useful. I might use it in my everyday life.

S1: I mean , it takes a lot of the energy of decision making first thing in the morning off your plate. I could see that being really useful. Definitely.

S6: Definitely. And I think for a lot of people , you know , young professionals coming into the workplace and not knowing how to dress for an office setting , for example , if someone were to create AI , that could help take those decisions off your plate and create like appropriate outfits for the workplace. That would be super useful for people.

S1: You know , a lot of discussions surrounding everyday use of AI involves how this this is all really kind of a work in progress and a little clunky. So did you find that to be the case with any of these decisions ? Definitely.

S6: A lot of the prompts took refining and it definitely people in my life who were better at using ChatGPT then I am. My brother reminded me that you can ask for different answers from ChatGPT. You can tell them you don't like what them. You can tell the AI that you don't like what it said , but it definitely takes a lot of refining to get the kind of responses that you want from it. Um , and also a lot of the outfits that style was generating generating for me were just silly. You know , I don't think we're at a point where we can plug things into a machine and magically have it take over everything. Darn.

S4: Darn.

S1: I mean , so give us your takeaways. Was this an effective way to run your life for a week ? I mean , I know you see some potential here , right ? Yeah.

S4:

S6: I don't think that we're at a point where I should be running anyone's life. I would just encourage anyone exploring ChatGPT and other forms of artificial intelligence to think of it as a tool. You know , paper planners are another tool that help me streamline my life. Um , but if someone handed me a paper planner full of things I should be doing , I would not listen because they don't know my life. I think that's the biggest failure right now is that , you know , ChatGPT is not a doctor , it's not a nutritionist , it's not a personal trainer. It can have conversations with you about what you need and why the meal that it gave you hurt your stomach. I think that we know our bodies best. We know our lifestyles best. And artificial intelligence , especially right now , is not able to to fill in that key element of of the uniqueness of every human being that's using it.

S4: I mean.

S1: Did you have any favorites in this experiment ? You know , I'm wondering if this process made you rethink how you approach something like food or fashion.

S6: So my favorite artificial intelligence that I used was actually Spotify's J. Its name is Xavier , and Xavier gives you music recommendations. Xavier comes in every few songs to switch up. The vibe tells you sort of the motivation behind that switch. A lot of the time it's , you know , songs that you were listening to at a certain point in your life or a certain genre. But I am a serial listener. I am not very good at going out and finding new music on my own. And so it was actually a really wonderful way to hear new music , discover new artists while still getting to hear my favorite songs and not having to go out and seek that out on my own. So I would definitely recommend trying out Xavier. And it's very it didn't feel intrusive into my life in the same way that other things did , which was really nice.

S1: That's pretty cool. I mean , would you recommend that anyone try this ? I mean , it seems like a pretty interesting way to structure your life , even on a short term basis , to just get familiar with artificial intelligence.

S6: I think that certainly it's worth experimenting with. It's worth. And putting some prompts. It's worth thinking about the ways that I could help you structure your life. So if you are struggling , I really struggle to plan for breakfast. I'm not very good at eating breakfast in the morning. And so the fact that every morning I had to get up and make the breakfast that I told me to make made a huge difference. Actually , I had more energy during the day , but I also found myself wishing that I hadn't committed quite so hard to doing exactly what it said , because sometimes I wanted an avocado on my omelet or sometimes I wanted cheese on my burger. And so I would encourage anyone who is experimenting with I and , you know , with these sort of plans that it can deliver to be flexible and to take them as a jumping off point and then sort of adapt according to what you need that day. So I think it's a good starting point.

S4: All right.

S1: Very interesting. I've been speaking with San Diego magazine associate editor , Amelia Rodriguez. Amelia , thank you so much for talking with us today. Absolutely.

S6: Absolutely. Thank you for having me on.

S1: With all of its uses. Are you worried I could take over humanity or are you excited about its potential ? Give us a call at (619) 452-0228. Leave a message or you can email us at midday at pbs.org. We love sharing your ideas right here on the show. And if you ever miss a show , you can find the Midday Edition podcast on all platforms. I'm Jade Hindman. Thanks for listening.